Projects

Ela, functional language

Ela is a modern programming language that runs on CLR and Mono. The language is dynamically (and strongly) typed and comes with a rich and extensible type system out of box. It provides an extensive support for the functional programming paradigm including but not limited to - first class ... [More] functions, first class currying and composition, list/array comprehensions, pattern matching, polymorphic variants, thunks, etc. It also provides some imperative programming features. Ela supports both strict and non-strict evaluation but is strict by default. The current language implementation is a light-weight and efficient interpreter written fully in C#. The interpreter was designed to be embeddable and has a clear and straightforward API. [Less]

compilervision

CompilerVision helps compiler developers to visualize, design, debug and optimize output code. Currently, the CompilerVision has tools that visualize specific data structures used in compiler development like control flow graphs, dominator trees, dominance frontiers and SSA intermediate forms. ... [More] Future versions of CompilerVision will have the following features: 1. Translation from any popular high-level languages like C, C++, Java or C++ to optimized assembly instructions for both CISC or RISC architectures like x86, ARM or MIPS (both CISC and RISC) 2. Virutal machines for architectures like ARM or MIPS. 3. Code optimizations based on various code improvement algorithms like loop optimizations, hoisting, aggressive dead code elimination, constant folding, copy propagation and pipelining and scheduling. [Less]

openpex

PEX (Provisioning and EXecution Environment) is a resource provisioning system based on VMs. PEX was constructed around the notion of using advance reservations as the primary method for allocating VM instances. A user, either through a web portal or through a web service can make a reservation for ... [More] any number of instances of a Virtual Machine that have to be started at a specific time and have to last for a specific duration. VMs are described by a template that is already registered in the system. [Less]

memx

To INSTALL or Use MemX now, go to Usage The MemX system, developed in the Operating Systems and Networks (OSNET) Lab at Binghamton University, provides a mechanism to virtualize the collective memory resources of machines in a cluster with zero modifications to your application or operating ... [More] system. Features: Auto-discovery and pooling of unused memory of nodes across physical or virtual machine networks. Fully transparent, linux kernel-space implementation (2.6.XX). No centralized servers. No userland components. (aside from the application). Completely transparent access to memory clients (no application or OS modifications). Live client state transfers (allows you to shutdown a client and reattach a block device on a new host or virtual machine). Sort of like a self-migrating block device without VMs Live server shutdowns. Allows you to migrate server memory of individual hosts without disconnecting clients. (Self-dispersing servers). MemX uses about 6000 lines of kernel code for all of the above features. How it Works: In MemX, a linux module is inserted into the vanilla kernel of every participating machine in the cluster, turning each machine into either a client or a server. The client module implements an optional block device. (The client can also be accessed from within the kernel). This device can be used either for mmap(), for swapping, for filesystems. It's your choice. Node Discovery: Servers announce (broadcast) themselves and their load statistics to every machine in the cluster at regular intervals. These announcements allow clients to make decisions for page allocation and retrieval across all available servers. Clients accordingly accept block I/O requests, either from the local linux swap-daemon or the file system layer, and service the I/O requests from cluster-wide memory. Advantages of our approach: We get the benefits of a kernel-space implementation, without changing the virtual memory system. The network protocol stack is bypassed because we don't need routing, fragmentation, nor a transport layer. Cluster-wide memory is managed and virtualized across a low-latency, high-bandwidth interconnect. Depending on the state of the cluster, any client or server can be made to join or leave the system at will. Any workstation with a vanilla linux kernel can be used to run any legacy memory-intensive application. Previous Work: The following is a list of other remote memory systems that are similar in spirit but do not satisfy the goals that we do: (Refer to "How it Works" for explanation of terms used) Classification Name Test Platform Used Code Available Test Cluster Size Test Network Used Page-Fault Time or speedup Main Caveat Self-migration capable Currently Active MemX Linux 2.6. Yes 140 GB DRAM total across Twelve 8-core 2.5ghz Opterons Gigabit Ethernet 80 usec Needs replication Yes Current JumboMem Linux/Unix Yes 250 Nodes, 4GB each, 2 Ghz Opterons 4X Infiniband 54 usec Application changes to malloc() library call No Current Nswap Linux 2.6. No Eight 512MB nodes, Pentium 4 Gigabit Ethernet unsure None No Current LamdaRAM unsure No unsure Wide-Area-Networks 80 ms? unsure No 2003? "SAMSON" Linux 2.2 No Seven 933mhz Pentiums Myrinet or Ethernet 300 usec Kernel Changes No 2003 "Network Ramdisk" Linux 2.0 or Digital Unix 4.0 Yes 233mhz Pentium or DEC-Alpha 3000 155 Mbps ATM or 10 Mbps Ethernet 10-20 msec User-level Servers No 1999 "User-level VM" Nemesis No 200mhz Pentium nodes 10 Mbps Ethernet Several millisec Application is changed No 1999 Berkeley "NOW" GLUnix + Solaris Yes 105 Nodes, UltraSparc I 1.28 Gbps Myrinet unsure unsure No 1998 User-level "Dodo" Linux 2.0, Condor-based No 14 Nodes, 200mhz Pentium 100 Mbps Ethernet speedup of 2 to 3 Application is changed No 1998 "Global Memory System" Digital Unix 4.0 No 5-20 Nodes, 225 MHz Dec Alphas 1.28 Gbps Myrinet 370 microseconds Kernel VM Subsystem Changes No 1999 "Reliable Remote Memory Pager" DEC OSF/1 No 16? Nodes, DEC Alpha 3000 10 Mbps Ethernet several millisec, speedup of 2 User-level Servers, not linux-based No 1996 Publications MemX: Supporting Large Memory Applications in Xen Virtual Machines, SlidesIn Proc. of November, 2007 Second International Workshop on Virtualization Technology in Distributed Computing (VTDC07). A workshop in cunjunction with Super Computing 2007, Reno, Nevada. M. Hines and K. Gopalan [Less]

linkedprocess

ANNOUNCEMENT: LoPSideD has been moved to the GitHub repository. The GoogleCode repository will exist until everything is stable at GitHub.The Linked Process project is focused on providing a specification and implementation of a distributed computing cloud that can be supported and leveraged by any ... [More] device with an Internet connection. A list of relevant links is provided below: A short Linked Process tutorial. A collection of Linked Process use cases. The Linked Process protocol specification. The LoPSideD Linked Process source, distributions, and javadoc. A short LoPSideD villein tutorial. A collection of various related materials. For those that wish to contribute to the Linked Process project, please be sure to join the project's associated discussion group. Finally, if you are running a Linked Process farm, please subscribe your countryside to lanl_countryside@lanl.linkedprocess.org so that your farm is recorded in the lanl_countryside registry. [Less]

civix

CIVIC is a modification to the Xen Hypervisor that allows you to over-subscribe a Virtual Machine's physical memory (DRAM) in such a way that it will have more DRAM than is available on a single physical host. Additionally, it is built to use network RAM to store pages that are not being ... [More] accessed instead of disk. CIVIC leverages the MemX system for the network RAM component while CIVIC itself implements the over-subscription capabilities (through hooks in shadow paging). CIVIC is a hypervisor-level solution only and does not require any kernel patches to the operating system, nor does it require the administrator to load kernel modules into the unmodified guest Virtual Machine. This means that you can use it with any operating system. Is it stable? Yeah, sort of. It's stable enough to benchmark, but not stable enough to use permanently. It can run HVM Linux and HVM OpenSolaris and Windows can make it up to the boot screen (until it crashes). Please contact the group members if you're interested. [Less]

estela

Estela is a Domain Specific Language for ML model implementation (and testing). It is an Object Oriented Language (but not Object Oriented Obsessed like Smalltalk). Estela's API is composed by two parts: small generic API (e.g. Generic types like Integer, Double, String, Date and Object) and a ... [More] extensive ML API (e.g. Classes like DataSet, Classifier, ClassifierEvaluation, Instance, and so on). Besides that, the language offers declarative constructions for some tedious tasks, and of course, a good array manipulation support. Estela is a dynamic (interpreted), strong typed, not fully Object Oriented language. A dynamic language is the best tool for experimental computing. They provide you with a simple way to load and unload data structures (e.g. Easily dynamic class loading). Estela Virtual Machine (VM) will run on top of a Java VM. No ML algorithm run in complexity time less than Ω(MN) in which M is the number of attributes (columns) and N is the number of records. The Estela Virtual Machine has lots of opportunities to explore parallelism (e.g. Threads running in a multi-core machine) processing and reduce complexity time to Ω(MN/p) in which p is the number of processor. Usage of threads and concurrent algorithms should be transparent to the user. [Less]

Managed Compiler Infrastructure

The Managed Compiler Infrastructure is a modern and intuitive compiler back end for managed languages.